1 /////////////////////////////////////////////////////////////////////// 2 // File: cjkpitch.h 3 // Description: Code to determine fixed pitchness and the pitch if fixed, 4 // for CJK text. 5 // Copyright 2011 Google Inc. All Rights Reserved. 6 // Author: takenaka@google.com (Hiroshi Takenaka) 7 // Created: Mon Jun 27 12:48:35 JST 2011 8 // 9 // Licensed under the Apache License, Version 2.0 (the "License"); 10 // you may not use this file except in compliance with the License. 11 // You may obtain a copy of the License at 12 // http://www.apache.org/licenses/LICENSE-2.0 13 // Unless required by applicable law or agreed to in writing, software 14 // distributed under the License is distributed on an "AS IS" BASIS, 15 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 16 // See the License for the specific language governing permissions and 17 // limitations under the License. 18 // 19 /////////////////////////////////////////////////////////////////////// 20 #ifndef CJKPITCH_H_ 21 #define CJKPITCH_H_ 22 23 #include "blobbox.h" 24 25 namespace tesseract { 26 27 // Function to test "fixed-pitchness" of the input text and estimating 28 // character pitch parameters for it, based on CJK fixed-pitch layout 29 // model. 30 // 31 // This function assumes that a fixed-pitch CJK text has following 32 // characteristics: 33 // 34 // - Most glyphs are designed to fit within the same sized square 35 // (imaginary body). Also they are aligned to the center of their 36 // imaginary bodies. 37 // - The imaginary body is always a regular rectangle. 38 // - There may be some extra space between character bodies 39 // (tracking). 40 // - There may be some extra space after punctuations. 41 // - The text is *not* space-delimited. Thus spaces are rare. 42 // - Character may consists of multiple unconnected blobs. 43 // 44 // And the function works in two passes. On pass 1, it looks for such 45 // "good" blobs that has the pitch same pitch on the both side and 46 // looks like a complete CJK character. Then estimates the character 47 // pitch for every row, based on those good blobs. If we couldn't find 48 // enough good blobs for a row, then the pitch is estimated from other 49 // rows with similar character height instead. 50 // 51 // Pass 2 is an iterative process to fit the blobs into fixed-pitch 52 // character cells. Once we have estimated the character pitch, blobs 53 // that are almost as large as the pitch can be considered to be 54 // complete characters. And once we know that some characters are 55 // complete characters, we can estimate the region occupied by its 56 // neighbors. And so on. 57 // 58 // We repeat the process until all ambiguities are resolved. Then make 59 // the final decision about fixed-pitchness of each row and compute 60 // pitch and spacing parameters. 61 // 62 // (If a row is considered to be proportional, pitch_decision for the 63 // row is set to PITCH_CORR_PROP and the later phase 64 // (i.e. Textord::to_spacing()) should determine its spacing 65 // parameters) 66 // 67 // This function doesn't provide all information required by 68 // fixed_pitch_words() and the rows need to be processed with 69 // make_prop_words() even if they are fixed pitched. 70 void compute_fixed_pitch_cjk(ICOORD page_tr, // top right 71 TO_BLOCK_LIST *port_blocks); // input list 72 73 } // namespace tesseract 74 75 #endif // CJKPITCH_H_ 76